Identification of cellular proteins upregulated by BALF

Knowledge of the mechanisms leading to adaptation of pathogens to their host environment and their interaction with host cells is crucial for understanding pathogenesis and for systematic development of life vaccines. One strategy is the identification of genes expressed in vivo. Several experimental approaches have been used recently to investigate which A. pleuropneumoniae genes are important in vivo, most importantly In Vivo Expression Technology (IVET), Signature-Tagged Mutagenesis (STM) and subtractive hybridization techniques. STM is based on comparison of pools of tagged transposon mutants recovered from infected animals to the pools prior to infection. Mutants that are not recovered likely have insertions in genes essential for survival; however, genes that only modulate virulence and genes that are involved in persistence are rarely identified by this approach (FULLER et al.

2000b; SHEEHAN et al. 2003). IVET allows the identification of promoters which are switched on in the host. An A. pleuropneumoniae IVET study was performed using a

riboflavin auxotrophic strain transformed with plasmids containing A. pleuro-pneumoniae genomic DNA fragments and promoterless riboflavin genes. These mutants only survive in the host if the promoter is active in vivo, leading to transcription of riboflavin synthesis genes. Promoters that are active in later phases of disease only and that might be involved in persistence would not facilitate survival of the mutants during the acute phase of infection. Therefore, such promoters cannot be identified by IVET. Various methods based on subtractive hybridization of cDNA have been described for detection of differentially transcribed genes.

Representational Difference Analysis (RDA) has been used previously to successfully identify A. pleuropneumoniae genes upregulated upon addition of BALF from day 7 post infection (BALTES 2002; BALTES et al. 2003a; BALTES et al.

2003b). An advanced subtractive hybridization technique, the Selective Capture of Transcribed Sequences (SCOTS) has been employed recently to identify A. pleuro-pneumoniae genes which are transcribed in necrotic porcine lung tissue (BALTES and GERLACH 2004). However, regulation of gene expression is not the only mechanism contributing to adaptation of pathogens to their host environment. Both the rate of translation and the protein stability or rate of degradation additionally influence the amount of a protein species in a cell. Therefore, mRNA levels don’t necessarily correlate with protein levels (ANDERSON and SEILHAMER 1997; GYGI et al. 1999). In addition, posttranslational modifications might affect the function of proteins, and even the subcellular localization might be of importance (DALLO et al.

2002; NOUWENS et al. 2003). Therefore, in addition to transcriptome analyses, the comparison of proteomes might yield valuable information about adaptive responses of pathogens to environmental changes (CASH 2003; NOUWENS et al. 2003;

CORDWELL 2004). In contrast to STM, IVET and most subtractive hybridization techniques, proteomic approaches, like 2D-PAGE combined with mass spectrometry, allow quantitative analysis of protein expression and modification. However, direct analysis of a pathogen’s proteome in infected tissue is extremely difficult due to the abundance of host proteins and would require separation of bacteria from host cells prior to analysis. A feasible alternative is the use of ex vivo models mimicking the host environment. Such a model, based on the addition of BALF to the culture medium, has been established in our group previously (TEUTENBERG-RIEDEL 1998; HENNIG et al. 1999) and was used successfully in an RDA study (BALTES 2002). Using BALF collected at different time points during infection, changes in the

host environment over the course of disease can be mimicked. Therefore, in order to complement and extend the results obtained by SCOTS and RDA, a proteomic approach, based on protein separation by 2D-PAGE and subsequent identification by mass spectrometry was used in this study. Proteins upregulated by the addition of BALF to the culture medium were to be identified, hypothesizing that these proteins would also be upregulated in vivo and might contribute to virulence.

Sample preparation and solubilization are crucial for satisfactory protein separation by 2D-GE. In order to separate proteins contained in the BALF from the bacterial cells, bacteria were pelleted by centrifugation before preparation of cell lysates.

Hence, whole cell lysates analyzed in this study do not include any secreted proteins, for example RTX toxins. Removal of RNA/DNA by enzymatic treatment, precipitation and washing steps were necessary to obtain sufficient resolution on 2D gels. It should be noted, that most likely, some proteins were lost during sample preparation steps due to incomplete precipitation or due to solubility in acetone during the washing steps aimed at reduction of the salt concentration. Conditions for IEF and SDS-PAGE were chosen to facilitate optimal separation and visualization of the majority of cellular proteins which showed a range of molecular mass from 30 kDa to 90 kDa with an isoelectric point between pH 4 and pH 7. Consequently, proteins with an isoelectric point below pH 4 and above pH 7, and very small or large proteins are not represented on the 2D gels in this study. Also, low abundance proteins and highly hydrophobic proteins tend to be underrepresented on 2D gels, thus differential expression of these proteins might not be detectable. Therefore, the proteins identified in this study only represent a subset of all proteins regulated by BALF.

Approximately 200 protein spots were visible on gels stained with Coomassie Brilliant Blue. At least 11 spots were found to be reproducibly upregulated by BALF. Four of the 9 protein spots analyzed by Q-TOF MS were identified as metabolic enzymes.

The finding that metabolic enzymes are regulated in a model mimicking the condition inside the host, is consistent with results from the RDA study (6 of 11 identified genes; BALTES 2002) and studies using IVET (4 of 7 identified genes; FULLER et al.

1999) or SCOTS (22 of 46 identified genes; BALTES and GERLACH 2004) to identify genes upregulated in vivo. Likewise, enzymes involved in metabolism were found to be necessary for virulence in two STM studies (10 of 20 and 21 of 55 identified genes, respectively; FULLER et al. 2000b; SHEEHAN et al. 2003). The

metabolic proteins upregulated by BALF identified in whole cell lysates were an aspartate ammonia-lyase (two spots; discussed in E.2), a dihydrolipoamide dehydrogenase and a phosphoribosylpyrophosphate (PRPP) synthetase. The dihydrolipoamide dehydrogenase showed upregulation by BALF taken at all three time points. In E. coli, this enzyme is a subunit of both the 2-ketoglutarate dehydrogenase and the pyruvate dehydrogenase; additionally it is involved in the glycine enzyme cleavage system (STEIERT et al. 1990). As part of the 2-ketoglutarate dehydrogenase it is involved in the conversion of 2-2-ketoglutarate to succinyl-CoA in the TCA cycle (GUEST and RUSSELL 1992). A transposon-insertion mutant in lpd coding for the dihydrolipoamide dehydrogenase was identified by Fuller et al. (2000b) in an STM study and found to be highly attenuated compared to wild type bacteria. This suggests that the dihydrolipoamide dehydrogenase is of importance for in vivo survival of A. pleuropneumoniae.

Expression of the PRPP synthetase was found to be upregulated only by BALF prior to infection and BALF from day 7 post infection; BALF from day 21 post infection led to increased spot intensity only in one of three gels. The enzyme catalyzes the generation of PRPP from ribose 5-phosphate, thereby diverting ribose from the energy generating pentose phosphate way to biosynthesis of pyridine nucleotide coenzymes, histidine, tryptophan and pyrimidine and purine nucleotides (CURTISS III et al. 1996). In E. coli activity of the PRPP synthetase is inhibited by binding of ADP to the ATP binding site (HOVE-JENSEN et al. 1986). Additionally, pyrimidine represses enzyme synthesis. There is no obvious role for a PRPP synthetase in A. pleuropneumoniae infection except for its role in biosynthesis; increased expression induced by BALF might be part of a more general metabolic adaptation to nutrient restriction in the host environment.

Two of the 9 proteins analyzed by Q-TOF MS showed no homology to any public data base entry; this is a higher proportion than obtained in STM studies (FULLER et al. 2000b; SHEEHAN et al. 2003; 2 of 20 and 2 of 55, respectively) and in SCOTS (3 of 46; BALTES and GERLACH 2004). Since at the time of analysis only a small portion of the A. pleuropneumoniae genome was sequenced and annotated in public databases, identification of proteins mainly depended on homology to annotated proteins of related species. Sequence information obtained by Q-TOF might not have been sufficient to allow identification of proteins with a relatively low similarity,

especially if the sequenced peptides derived from variable rather than conserved regions within a protein. The progress in sequencing of the A. pleuropneumoniae genome in combination with repetition of mass spectroscopy to obtain further peptide sequences should allow identification of these proteins in the future.

E.2. Characterization of the A. pleuropneumoniae aspartate ammonia-lyase

Among the proteins identified to be upregulated in whole cell lysates the aspartate ammonia-lyase was further investigated for two reasons. (i) Aspartase, like the DMSO-reductase previously identified (BALTES et al. 2003a), is involved in anaerobic respiration which might be important for persistence of A. pleuro-pneumoniae in the respiratory tract, and (ii) the aspartase of H. influenzae had been shown to possess a plasminogen-binding activity which might be related to virulence (SJOSTROM et al. 1997).

The initial observation, that the aspartase protein is upregulated by addition of BALF was confirmed by an enzymatic assay, showing increased aspartase activity in cultures treated with BALF and anaerobic cultures; the relative increase of aspartase activity upon anaerobic culture was twofold in comparison to the aerobic control culture, and induction due to BALF was slightly less than that.

The highly increased aspartase activity in an E. coli strain harboring the A. pleuropneumoniae aspA ORF demonstrated that aspA indeed codes for a functional aspartase. Furthermore, it was noteworthy, that the E. coli strain carrying the expression vector only showed no detectable aspartase activity. This finding is concordant with observations of JERLSTRÖM et al. (1987) and GUEST and RUSSELL (1992) that E. coli aspartase is not expressed under standard conditions.

The situation in A. pleuropneumoniae differs from this, as aspartase activity was clearly detectable under standard culture conditions.

Using Western blot analysis and aspartase assays, the A. pleuropneumoniae aspartase was found to be localized in the cytoplasm, as expected. Absence of aspartase activity in any cellular fraction obtained from chloroform preparation of periplasmic proteins could be due to negative effects of chloroform on aspartase function.

The construction of an isogenic aspA in-frame mutation and its complementation in combination with the investigation of aspartase activity in the respective strains clearly showed that A. pleuropneumoniae AP76 has a single aspA gene. Using aspA specific primers, a PCR product of the correct size could be amplified from all serotype reference strains, suggesting that the genetic basis for aspartase activity is not serotype specific. Based on these results, the expression of an active aspartase in the reference strains was not evaluated, as it would have been outside the scope of this study.

The determination of the transcriptional start site and the presence of a putative FNR binding domain centered at position -42.5 implied that A. pleuropneumoniae aspA - like the E. coli aspA - possesses a class II FNR-dependent promoter (JERLSTRÖM et al. 1987; WOODS S.A. and GUEST 1987), and thus, FNR would enhance transcription of aspartase. This hypothesis was supported by the finding that both the amount of aspartase and the aspartase activity in anaerobically grown bacteria are increased compared to aerobic conditions. Furthermore, no aspA-specific cDNA could be transcribed from mRNA of A. pleuropneumoniae AP76 grown under aerobic conditions.

In order to further support the transcriptional activation of aspA under anaerobic conditions and upon the induction of BALF, an A. pleuropneumoniae ∆aspA::luxAB strain (a mutant carrying a transcriptional aspA-luxAB fusion on the chromosome) was constructed. The luciferase of Photorhabdus luminescens was chosen as it has been used previously to perform gene expression studies in bacteria due to its stability at 37°C and its short half-life (WINSON et al. 1998; FRANCIS et al. 2000). In this mutant, no detrimental influences due to copy number are to be expected as they can occur in plasmid-based systems, which have been used previously to investigate promoter function in A. pleuropneumoniae (BOEKEMA et al. 2004). Further, the fusion resulted in an artificial aspA-luxAB-operon structured similar to the A. pleuropneumoniae urease (BOSSE and MACINNES 1997) and tbpBA operons (TONPITAK et al. 2000), with the stop codon overlapping the Shine-Dalgarno consensus sequence of the downstream ORF. This structure was considered to minimize potential polar effects. Using this mutant, a clear-cut transcriptional activation upon anaerobic growth and upon the addition of BALF was observed without the high background seen in aerobic control cultures in the aspartase assay.

These findings imply, that transcription of aspA is upregulated by anaerobiosis and suggest that one component regulating the A. pleuropneumoniae aspartase might be HlyX, the A. pleuropneumoniae FNR-homologue (GREEN and BALDWIN 1997). This hypothesis is strongly supported by the observation, that upregulation of aspartase activity under anaerobic conditions is abolished in an HlyX-negative mutant.

Therefore, the regulation would resemble the situation in E. coli, where aspA transcription is likewise upregulated by FNR under anaerobic conditions (JERLSTRÖM et al. 1987; WOODS S.A. and GUEST 1987).

However, as induction of aspartase activity in A. pleuropneumoniae wt and luciferase activity in A. pleuropneumoniae ∆aspA::luxAB was shown upon induction with BALF during aerobic growth with shaking, it appears unlikely that a decreased oxygen tension alone is responsible for the upregulation of aspartase expression. The suggested influence of other factors coregulating the expression of HlyX dependent genes is supported by the finding of SOLTES and MACINNES (1994), who showed that HlyX-dependent expression of an frdA-lacZ fusion in E. coli varied depending on growth phase and carbon source. Therefore, we hypothesize that as of yet unknown factors in the BALF are responsible for upregulation of aspartase activity and transcription of aspA. One possibility would be that these factors have an HlyX-mediated effect on AspA expression. Further, since the putative FNR-binding site upstream of aspA (GTGAT-CTAA-ATCAC) also shows high homology with the E. coli CRP (cAMP receptor protein)-binding site (EBRIGHT et al. 1989; AAAT-GTGAT-CTAG-ATCAC-ATTT), regulation by CRP would also seem to be a possibility.

However, nothing is known about CRP-homologues in A. pleuropneumoniae, and also it cannot be excluded that other transcriptional regulators and promoter structures are involved. Therefore, further studies are needed to elucidate the possible role of HlyX in aspartase regulation and as a global regulator in A. pleuropneumoniae.

The reduced growth observed for A. pleuropneumoniae ∆aspA under anaerobic conditions and the lack of detectable plasminogen-binding activity, as it had been described for H. influenzae (SJOSTROM et al. 1997), led to the hypothesis, that the major function of aspartase in A. pleuropneumoniae virulence might be the production of fumarate acting as electron acceptor for anaerobic respiration as described for E. coli (JENNINGS and BEACHAM 1993). This hypothesis was

supported by the finding that growth of the mutant was not impaired under aerobic conditions, thereby implying that the role of aspartase in amino acid metabolism is unlikely to be the cause for reduced growth under anaerobic conditions. Since alternative anaerobic respiration pathways are likely to compensate for each other's absence in the presence of suitable substrates, the second A. pleuropneumoniae pathway known for anaerobic respiration which is driven by the DMSO reductase, was deleted by constructing the double mutant A. pleuropneumoniae ∆aspA∆dmsA.

The finding, that the growth of this double mutant was indistinguishable from that of the single mutant A. pleuropneumoniae ∆aspA, and our observation, that the lack of the dmsA gene alone does not diminish growth under anaerobic conditions in vitro, suggest that aspA but not dmsA is important for anaerobic growth in vitro.

The challenge of pigs with A. pleuropneumoniae ∆aspA led to clinical disease with only slightly lower clinical score than observed in pigs challenged with A. pleuropneumoniae wt. The absence of dyspnea from day 3 onwards observed in pigs challenged with A. pleuropneumoniae ∆aspA is most likely due to the lack of animals with very severe lung lesions as they occurred upon infection with A. pleuropneumoniae wt. Thus, although the lung lesion score did not differ significantly, it was reduced by almost half in pigs challenged with A. pleuro-pneumoniae ∆aspA.

The clinical score (based on dyspnea, vomitus and coughing) of pigs challenged with A. pleuropneumoniae ∆aspA∆dmsA was significantly reduced in comparison to the groups challenged with A. pleuropneumoniae wt or A. pleuropneumoniae ∆aspA. As clinical symptoms are primarily due to a general colonization of the airways rather than the presence of sequestered tissue, this observation supports the relevance of our finding, that A. pleuropneumoniae ∆aspA∆dmsA could only very sporadically be reisolated from intact lung tissue on days 7 and 21 post infection. Together, these findings might suggest a role for enzymes involved in anaerobic respiration in colonization of the respiratory epithelium.

The lung score of pigs infected with either one of the mutants was decreased by 50%

with respect to the wild type (although the challenge dose for both mutants was approximately double that of the wild type). Despite the lack of statistical significance, this finding is noteworthy, as it strongly supports the attenuation of both strains.

Further, it implies that A. pleuropneumoniae has additional metabolic pathways to

grow and persist within necrotic lesions. Possibly, other enzymes facilitating anaerobic respiration such as the nitrate/TMAO reductase (NZ_AACK01000004.1) might play a role as it has been described for Mycobacterium bovis BCG infections (FRITZ et al. 2002; WEBER et al. 2000). One possible enzyme contributing to anaerobic respiration could be the homologue to a Pasteurella multocida periplasmic nitrate reductase (acc. no. AAK03681) identified by SHEEHAN et al. (2003) in an A. pleuropneumoniae STM study.

Based on the results of this study, we conclude that enzymes involved in anaerobic respiration play a role in A. pleuropneumoniae persistence and virulence. They appear to be important not only for survival in necrotic lesions but, surprisingly, might be required for long term colonization of intact respiratory epithelium in a presumably aerobic environment.

E.3. Role of Fur as global gene regulator in A. pleuropneumoniae